A reviewer who tested AI Siri in Apple's iOS 27 developer beta started the summer convinced the assistant would become the iPhone's tool for everything, and finished it barely using the thing. The assistant had not got worse over those weeks. It had got noticeably better, and it still lost. What it lost to was not a rival chatbot but the reviewer's own thumbs: Instagram again, aimless scrolling through Yelp, typing queries by hand instead of asking.
The first impression was strong. Asked to give a tour of San Francisco, the assistant returned answers reliable enough to use and more useful than what earlier versions of Siri produced, wrapped in a chatbot app that looked tidy and recognisably Apple's. Over the following month in the beta the standout capability turned out to be personal context: the assistant indexes the contents of the phone, which makes digging out an old message or photo easier than it has ever been.
Then usage fell off, and the reviewer's account of why is unusually honest. The disappointment was sharper precisely because the assistant had improved. It did not change how he interacts with the phone the way he had hoped it would, and the hope was set high by the two things Apple can claim here that most rivals cannot: the assistant is built directly into the device, and Apple pays more attention to privacy than many companies working on AI. He was content enough with Siri turning into, effectively, the phone's search bar. Interest never curdled into irritation over the summer. It just went flat. It also helped that Claude, from Anthropic, was already on the phone, and he was used to it and reached for it often enough.
The analysts quoted around this agree that none of it costs Apple much. Dipanjan Chatterjee, vice-president and principal analyst at Forrester, expects Siri to be criticised for trailing some competitors but doubts that will send large numbers of iPhone owners looking for an alternative; a segment of the base trusts Apple with their data more than they trust any other company, and AI Siri stands to inherit that trust. Ramon Llamas, research director for mobile devices at International Data Corporation, makes the same bet from the other direction: users deeply embedded in Apple's ecosystem who had doubts about the safety of other AI tools may be willing to try this one.
And the reviewer may simply be an outlier. Other testers are still using the assistant heavily. Avi Greengart, founder and lead analyst at Techsponential, says he never trusted Siri enough to hand it anything useful before; in the latest beta he goes to it more often and for more things. It finds information buried in the phone faster and more effectively than any other method, its general knowledge is significantly better, and dictation has improved too.
Notice what the defence of Siri is not. Neither analyst argues the assistant is better than the alternatives. The case rests on trust and on inertia — reasons not to leave rather than reasons to use. That is a solid business position and a weak product position, and what it describes is a feature that succeeds by not being switched away from. Apple has spent a long time being late to this; the argument that being late does not matter is, read closely, an argument that the feature did not have to be very good.
The more interesting thing is that the reviewer buried his own finding. The one capability that kept working was personal context — search across his own messages and photos — and he frames it as the assistant being demoted to a search bar. That is the single part of this product no competitor can build, because no competitor has the index. Being the search layer over everything a person has already said, photographed and saved is not a consolation prize. It is the one job Apple is uniquely placed to do, described here as a letdown because it was not the tool for everything.
What nobody in the piece takes up is the name. The review notes that a significant share of iPhone owners may refuse the new Siri precisely because AI is in its name and in the device, whether out of moral objection or a broader distrust of AI tools, and that better features will not move them. Left alone is what that means for a company whose assistant arrives by default on hardware people have already bought. Apple's competitors sell AI to people who went looking for it. Apple installs it on the phones of people who did not.
So the beta produces two failure modes at once, and neither is about capability. The users who object to AI will not turn it on. The users who like AI already have something else open. Between them sits the reviewer, who liked it, used it less each week, and concluded the problem was not the assistant but himself — the most useful thing Apple could have learned this summer, and the least likely to survive contact with an engineering roadmap.